JP2001111816A - Document image processing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve problems of a conventional document image processing device where a described character string and a seal stamp in a document image have been hard to see or have been illegible because of a hindrance due to a pattern crowded with very small noises caused by optical scanning of a document and much communication cost and storage cost have been required due to a huge image data amount because the noises consist of very small dot patterns and linear patterns. SOLUTION: The document image processing device is provided with a means that detects presence of noises consisting of very small patterns and a means that eliminates the noise from a captured document image to clear up the image.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for processing a document image, and more particularly, to a method for processing a document image for detecting and removing a subtle change in a pattern printed on a document.

[0002]

2. Description of the Related Art JP-A-6-215177 "High-speed document confirmation system and document for confirmation and document confirmation method" (applicant, Sigma Imaging Systems, Inc.) and JP-A 7-1879 "guarantee documents" (applicant, Traditionally, as described in The Standard Register Company, guarantee documents, such as various certificates, checks, shares and bonds, are valid documents and are not counterfeit or photocopied. Often, a message that was not visible in the original document, such as the message "copy", appears in a copy of the document, thereby preventing the use of a fake document. Have been. In this specification, the appearing partial image of the message is called a message pattern. On these documents, for example, patterns having lines with different resolutions are printed, or patterns formed with inks having different densities or fluorescent inks are printed. A document confirmation system has been proposed in which a surface image collected by a scanner is displayed on a screen, and an operator visually confirms whether or not a message pattern appears in the document.

[0003] Similarly, Japanese Patent Application Laid-Open No. Hei 6-215177 discloses an image in a predetermined area in order to confirm the validity of a document such as a check.
It is described to automatically detect whether the "LID" appears by optical scanning.

[0004]

When a message pattern appearing by optical scanning overlaps a written character string or stamp, the message pattern appearing by optical scanning becomes an obstacle, and the written character string or stamp is difficult to see. Or, there was a problem that it could not be read.

[0005] Further, since the message pattern is formed from a minute point-like pattern or a linear pattern, the data amount of a document image in which the message pattern appears cannot be reduced by a known standard compression method. The amount of image data becomes enormous. For this reason, when the document image is communicated to a remote place, there is a problem that communication cost is required.
Further, when storing the document image, there is a problem that the amount of image data becomes enormous and occupies a large storage capacity.

Further, Japanese Patent Application Laid-Open No. Hei 6-215177 discloses a method for automatically checking whether a message appears by optical scanning by comparing images of a predetermined area in order to confirm the validity of a document such as a check. Is described. However, automatic detection of a message in which a message appears at an arbitrary position in the document is not considered.

It is a first object of the present invention to remove a message pattern that has appeared by optical scanning of a document and to clarify the document image. This makes it easier to read the written character string or the seal on a document in which the message pattern overlaps the character string or the seal written on the document. Further, since the message pattern is removed, the data amount of the document image can be reduced, and the cost of storing and communicating the document image can be reduced.

[0008] A second object of the present invention is to remove a message pattern generated by optical scanning of a document and to add a known electronic watermark to electronic image data obtained by clarifying the document image. It is. By adding document information such as the source and history of the document, the contents of the description, and the removed message pattern to the clarified image data as a digital watermark, the security of document storage and communication can be improved. Known methods of digital watermarking include:
Ryoichi Sasaki, "Introduction to Internet Security" 19
The digital watermarking technology described on pages 7 to 200 (Iwanami Shinsho, issued March 19, 1999) may be used.

A third object of the present invention is to remove a point noise and a line noise from a message pattern formed of a minute pattern, thereby removing the message pattern and clarifying a document image. Is to provide.

A fourth object of the present invention is to provide a message having a message appearing at an arbitrary position, not a fixed position in the document, as a message pattern formed from a minute pattern by optical scanning of the document. That is, whether or not noise is detected. This makes it possible to determine whether the document to be optically scanned is a document copied by a copying machine or the like.

[0011] A fifth object of the present invention is to reduce the amount of image data by image clarification for removing minute patterns, and after image clarification processing, transmit the image data to a remote place via a network. As a result, communication costs and storage costs for image data can be reduced.

[0012]

In order to achieve the first object, first means for detecting the presence or absence of noise composed of minute patterns, and elimination of the noise from a sampled document image to clarify the image And a second means for clarifying an image only when noise composed of minute patterns is detected by the first means.

[0013] To achieve the second object, a first means for detecting the presence or absence of noise composed of a minute pattern;
A second means for removing the noise to clarify the image and a third means for adding a digital watermark to the clarified image based on the document information are provided.

In order to achieve the third object, means for removing noise and clarifying an image include means for removing noise in the form of isolated points and means for removing noise in the form of horizontal and vertical line segments. Means for removing noise in the shape of a line segment in an oblique direction.

In order to achieve the fourth object, the means for detecting the presence or absence of noise composed of minute patterns includes means for generating black runs and means for counting the number of black runs. Alternatively, the means for detecting the presence / absence of noise composed of minute patterns includes means for inputting a low-resolution image, means for inputting a high-resolution image, and means for comparing the low-resolution image with the high-resolution image. Is provided.

In order to achieve the fifth object, means for detecting the presence or absence of noise composed of minute patterns, means for removing the noise to clarify an image, and means for clarifying the document information with respect to the clarified image A means for giving a digital watermark based on the image data, transmitting the clarified image with a reduced data amount to a remote place via a network, and storing the clarified image with the digital watermark in a mass storage device .

[0017]

FIG. 1 is a block diagram of a document image processing apparatus according to an embodiment of the present invention. FIG. 2 is an example of a target document image. Document image 2 obtained by optical scanning
At 00, a message that was not visible in the original document, such as the message "copy", appears in an optical scan of the document. The appearing partial images 201, 202, and 203 of the message are called message patterns. Description character strings 210, 211, 212
When these message patterns overlap with each other, there is a problem that the written character string is difficult to see and difficult to read. Further, the message patterns 201, 202, and 203 are formed of minute patterns, and point noises and linear noises are concentrated. Since such noises are dense in the document image, if a known standard data compression method, for example, the MMR compression method (Modified, Modified Read) is applied to the document image, There is no effect of data compression, and there is a problem that the data amount is rather increased. For this reason, it is necessary to remove the noise to clarify the document image. The surface of the document is optically scanned by the photoelectric conversion unit 100 to collect a document image. For the document image, the noise presence / absence detection unit 101 detects a dense portion of minute noise, that is, the presence / absence of a message pattern. When it is determined that there is noise, the image clarifying unit 102 removes the noise. FIG. 3 is an example of a clarified document image. An image 300 is a result image of the document image 200 that has been subjected to a clearing process for removing noise in the image clearing unit 102. In the image 300 after clarification, message patterns 201, 202, and 2 composed of minute noises generated by optical scanning.
03 is removed, and only the description character strings 310, 311 and 312 appear. For this reason, the data amount of the image after clarification can be reduced. The transmission unit 103 converts the clarified image into a known standard data compression method, for example,
After applying the MMR compression method (Modified Modified Read method) to reduce the amount of data, the network 10
4 to the remote location, and the remote receiving unit 105
To receive. The digital watermark adding unit 106 adds a known electronic watermark to the image after the clarification based on document information, for example, information such as a source, a history, and a removed message pattern. Then, the clarified image data is stored in the storage unit 107 including a large-capacity storage device.

Message patterns 201, 202, 20
Numeral 3 is formed from a minute pattern, and point noise and linear noise are dense. FIG. 14 is an enlarged view of a part of the message pattern. A part 1400 of the message pattern includes isolated point noises 1401, 1402, and 1414.
03 is crowded. FIG. 9 is an enlarged view of a part of a message pattern in which line segment noise in the horizontal direction is dense. Line segment noises 901, 902, and 903 in the horizontal direction are concentrated in a part 900 of the message pattern. FIG. 11 is an enlarged view of a part of a message pattern in which vertical line segment noise is dense. A vertical line segment noise 1101 includes a part 1100 of the message pattern.
1102, 1103, and 1104 are dense. FIG.
FIG. 4 is an enlarged view of a part of a message pattern in which line segment noise in an oblique direction is dense. Part 1 of message pattern
200 includes oblique line segment noises 1201 and 1202;
1203 are dense. Depending on the document, any of these point noise or linear noise may appear as a message pattern, or may appear in combination with point noise or linear noise.It is not possible to specify the type of noise in advance. Can not.

FIG. 4 is a view for explaining the processing steps of the noise detection unit 101. In step 400, an image having a normal resolution is input, and in step 401, the image is converted into a monochrome binary image. Then, in step 402, a black line segment (referred to as a black run) composed of the horizontal connection of the black pixels is generated. Then, in step 403, the number of the black runs is counted, and the density of the black runs is calculated. Step 404
To determine the presence or absence of noise. In the presence / absence determination, for example, if the density of the black run is greater than a predetermined value, it is determined that there is a message pattern.

FIG. 5 is a diagram for explaining the processing steps of the image clearing unit 102. Either point noise or linear noise may appear as a message pattern, or point noise or linear noise may appear in combination, and the type of noise cannot be specified in advance. For this reason,
In step 500, isolated point noise is removed. And
In step 501, horizontal linear noise is removed. Next, in step 502, the black run is rotated 90 degrees, and in step 5
At 03, vertical line noise is removed. Since the black run is rotated by 90 degrees, the same processing method may be used for removing the horizontal linear noise in step 501 and removing the vertical linear noise in step 503. Next, in step 504, the black run is rotated 90 degrees to return to the original erect image. Then, Step 50
In step 5, oblique linear noise is removed.

FIG. 6 is a diagram for explaining the processing steps of another embodiment of the noise detection unit 101. In the optical scanning of the document surface, a low-resolution image is input in step 600, and the image is binarized in step 602. On the other hand, a high-resolution image is input in step 601, and the image is binarized in step 603. In a low-resolution image, a minute pattern change is not sampled, so that a message pattern composed of point-like or linear noise does not appear.
On the other hand, in a high-resolution image, a minute pattern change is sampled, so that a message pattern composed of point-like or linear noise appears. Using this principle, step 604
In, an image difference between a low-resolution binary image and a high-resolution binary image is extracted. In step 605, the magnitude of the image difference value is compared to determine the presence or absence. If the image difference value is large, it is determined that a message pattern has appeared by optical scanning. On the other hand, if the image difference value is small, it is determined that the message pattern has not appeared by optical scanning. In a paper document copied by a copying machine or the like, a message pattern appears at the time of copying. Therefore, since both the low-resolution binary image and the high-resolution binary image include the copied message pattern, the value of the image difference in step 604 is small, and the message pattern has not appeared by optical scanning. Is determined. This makes it possible to determine whether the input document is a copied document or an original document.

FIG. 7 is a diagram for explaining the configuration of a document processing apparatus for judging whether a document to be optically scanned is a document copied by a copying machine or the like or an original document.
A photoelectric conversion unit 700 collects a surface image of the document, and a noise presence / absence detection unit 701 detects the presence / absence of a message pattern. The noise presence / absence detector 701 may have the configuration shown in FIG. The result output part 7 detects the presence or absence of the message pattern.
02 is output as a warning sentence or warning sound on the screen.

FIG. 8 shows another embodiment of the present invention. The surface of the document is optically scanned by the photoelectric conversion unit 800 to collect a document image. Noise detection unit 801 for the document image
, A dense portion of minute noise, that is, the presence or absence of a message pattern is detected. If it is determined that there is noise, the image clarification unit 802 removes the noise. The digital watermark adding unit 803 adds a known electronic watermark to the image after the clarification based on document information, for example, information such as a source, a history, and a removed message pattern.

FIG. 10 is a diagram for explaining a method of removing the horizontal line segment noise shown in step 501. For the image represented by black runs, the black run of interest is now 1
000, the starting point coordinates XS0, 1010 of the target black run on the scanning line above the target black run 1000
It is determined whether a white area 1001 exists near the end point coordinates XE0 and 1011. Similarly, on the lower scanning line of the target black run 1000, the start point coordinates XS0 and 1010 and the end point coordinates XE0 and 101 of the target black run.
It is determined whether or not a white area 1002 exists in the vicinity of “1”. If the white areas 1001 and 1002 exist on the scanning lines above and below the black run 1000 of interest, the black run is deleted using the black run as line segment noise.

FIG. 13 is a view for explaining a method of removing the line segment noise in the oblique direction shown in step 505. For the image represented by black runs, the black run of interest is now 1
Assuming that the start coordinate XS0, 1310 of the target black run on the scanning line above the target black run 1300,
And the end point coordinates XE0 and 1311, the start point coordinate 13
30, it is determined whether there is a black run 1301 whose end point coordinates 1331 have a predetermined positional relationship. Similarly, it is determined whether there is a black run 1302 having a predetermined positional relationship between the start point coordinates 1320 and the end point coordinates 1321 on the scan line below the target black run 1300. If the black runs 1301 and 1302 exist on the scanning lines above and below the black run 1300 of interest, the black run 1330 is deleted using the black run as oblique line segment noise.

[0026]

According to the present invention, a written character string or a seal can be easily read in a document in which a message pattern overlaps a character string or a seal described in the document. Further, since the message pattern is removed, the data amount of the document image can be reduced, and the cost of storing and communicating the document image can be reduced.

Further, according to the present invention, the clarified document image data from which the message pattern has been removed is clarified by using document information such as the source, history, description contents and the removed message pattern of the document as an electronic watermark. By adding the document image data to the stored document image data, the security of document storage and communication can be improved.

Further, according to the present invention, it is possible to determine whether the document to be optically scanned is a document copied by a copying machine or the like.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a document image processing apparatus according to an embodiment of the present invention.

FIG. 2 is an example of a target document image.

FIG. 3 is an example of a clarified document image.

FIG. 4 is a diagram illustrating a processing process of a noise presence / absence detection unit 101.

FIG. 5 is a diagram illustrating a processing process of an image clarifying unit 102.

FIG. 6 is a diagram illustrating a processing process of another embodiment of the noise presence / absence detection unit 101.

FIG. 7 is a diagram illustrating a configuration of a document processing apparatus that determines whether the document to be optically scanned is a document copied by a copying machine or the like or an original document.

FIG. 8 is another embodiment of the present invention.

FIG. 9 is an enlarged view of a part of a message pattern in which horizontal line segment noise is dense.

FIG. 10 is a diagram illustrating a method for removing line segment noise in the horizontal direction shown in step 501.

FIG. 11 is an enlarged view of a part of a message pattern in which vertical line segment noise is dense.

FIG. 12 is an enlarged view of a part of a message pattern in which line segment noise in a diagonal direction is dense.

FIG. 13 is a diagram illustrating a method of removing line segment noise in an oblique direction shown in step 505.

FIG. 14 is an enlarged view of a part of a message pattern.

[Explanation of symbols]

101: Noise detection unit, 102: Image clarification unit, 1
06: digital watermarking unit, 200: document image, 201:
Message pattern, 300 ... image after clarification, 403 ...
Black run number counting step, 500 point-like noise removal step, 501 horizontal line noise removal step, 600 low-resolution image input step, 601 high-resolution image input step, 604 image difference step, 901 horizontal line segment Noise, 1101... Vertical segment noise, 1203
... Oblique line direction line segment noise, 1401... Isolated point noise.

 ────────────────────────────────────────────────── ─── Continued on the front page F term (reference) 5B057 AA11 BA02 BA29 CA02 CA06 CA12 CA16 CB02 CB06 CB12 CB16 CC02 CD04 CD05 CE02 CG07 CH08 DA08 DA17 DB02 DB05 DB08 DC02 DC32 DC36 5C076 AA02 AA14 AA21 AA22 AA40 BA06

Claims (6)

[Claims] 1. A reading means for reading a text image, and inputting the text image read by the reading means,
A first detecting means for detecting an image pattern in the text image and outputting the detection result; and inputting the text image read by the reading means,
A second detection unit that detects an image pattern in the text image at a higher resolution than the first detection unit and outputs the detection result; and inputs a detection result of the first detection unit and binarizes the input. Binarizing means for outputting the detected result of the first detecting means, inputting the detected result of the second detecting means, and outputting the binarized detection result of the second detecting means; Image difference means for inputting the detection result of the first detection means that has been converted into a value and the detection result of the second detection means having been converted to a binary value, and outputting a difference result between the two images; And noise detection means for detecting noise from the difference result, image clarification means for inputting the text image and the noise, and outputting a clarified image after removing the noise from the text image, A sentence image processing device comprising: 2. The text image processing apparatus according to claim 1, further comprising electronic watermarking means for inputting the clarified image, adding a digital watermark to the clarified image, and outputting the digitalized watermark. Text image processing device. 3. The text image processing apparatus according to claim 1, wherein said image clearing means has means for removing point-shaped noise and means for removing linear noise. Sentence image processing device. 4. The text image processing apparatus according to claim 1, wherein said binarizing means performs black and white binarization. 5. The text image processing apparatus according to claim 4, wherein the noise presence / absence detecting means generates a connection of black pixels included in the difference result, and calculates the number or density of the connection. A sentence image processing apparatus characterized by determining the presence or absence of noise. 6. The text image processing apparatus according to claim 1, further comprising a transmission unit that inputs the clarified image, converts the input image into transmission data, and transmits the transmission data. Text image processing device.